Integrative cross tissue analysis of gene expression identifies novel type 2 diabetes genes

Abstract

14To understand the mechanistic underpinnings of type 2 diabetes (T2D) loci mapped through GWAS, we 15 performed a tissue-specific gene association study in a cohort of over 100K individuals (n cases ⇡ 26K, 16 n controls ⇡ 84K) across 44 human tissues using MetaXcan, a summary statistics extension of PrediXcan. 17We found that 90 genes significantly (FDR < 0.05) associated with T2D, of which 24 are previously 18 reported T2D genes, 29 are novel in established T2D loci, and 37 are novel genes in novel loci. Of these, 1913 reported genes, 15 novel genes in known loci, and 6 genes in novel loci replicated (FDR rep < 0.05) in an 20 independent study (n cases ⇡ 10K, n controls ⇡ 62K). We also found enrichment of significant associations 21 in expected tissues such as liver, pancreas, adipose, and muscle but also in tibial nerve, fibroblasts, and 22 breast. Finally, we found that monogenic diabetes genes are enriched in T2D genes from our analysis 23 suggesting that moderate alterations in monogenic (severe) diabetes genes may promote milder and later 24 onset type 2 diabetes. 25. CC-BY 4.0 International license peer-reviewed) is the author/funder. It is made available under a The copyright holder for this preprint (which was not . http://dx.doi.org/10.1101/108134 doi: bioRxiv preprint first posted online Feb. 13, 2017; 2 Introduction 26Type 2 diabetes (T2D) is a complex disease characterized by impaired glucose homeostasis resulting from 27 dysfunction in insulin-secreting pancreatic islets and decreased insulin sensitivity in peripheral tissues 28 [1]. In addition to environmental factors such as a sedentary lifestyle and poor diet, genetic susceptibility 29 is an important contributor to the development of T2D [2]. Genome-wide association studies (GWAS) 30 have uncovered more than 100 loci that significantly associate with either T2D or glucose-related traits 31 [3, 4, 2]. However, the majority of single nucleotide polymorphisms (SNPs) significantly associated with 32 T2D reside in intronic and intergenic regions rather than protein-encoding regions [5, 6]. The results from 33GWAS suggest an important role for genetic variation that regulates gene expression rather than altering 34 codon sequence [7] and have motivated e↵orts to map the regulatory landscape of the genome [8, 9, 10]. 35Indeed, sets of trait-associated SNPs are enriched for variants that associate with gene expression (i.e. 36expression quantitative trait loci or eQTLs) [11] and that occupy DNAse hypersensitivity sites (DHS) [12] 37 -regions overrepresented for eQTLs per se [13]. Moreover, DHS explain a disproportionately high share 38 of SNP heritability [14] across 11 complex traits [15] and eQTLs mapped in insulin-responsive peripheral 39 tissues similarly "concentrate" SNP heritability estimates for T2D [16]. human hepatocytes [17]. Moreover, Sort1 knockdown and overexpression studies in mice altered LDL-C 46 and very low density lipoprotein (VLDL) levels [17]. In a study of the FTO locus harboring the strongest 47 associati...

A statistical framework for cross-tissue transcriptome-wide association analysis

“…Even when two tissues do not share eQTL, associations in the non-causal tissue still frequently passed the significance threshold, most likely due to LD between eQTL. These results are consistent with our experience and discussions in the literature 20,72 . We also note that these issues may become even more complex when sample sizes and imputation power vary across tissues.…”

Genetic discovery and translational decision support from exome sequencing of 20,791 type 2 diabetes cases and 24,440 controls from five ancestries

“…Twenty genes with T2D-associated transcript levels . We selected genes with significant associations in a pre-publication 52 tissue-wide T2D association analysis (i.e. testing for association between the genetic component of tissue-level gene expression and T2D), with associations considered significant if they survived Bonferroni correction for all tested genes and all tested tissues.…”

“…Next, we investigated whether effector genes that mediate GWAS associations – which mostly correspond to variants of uncertain regulatory effects – were also enriched for coding variant gene-level associations. We tested for associations within two sets of predicted effector genes: a curated list of 11 genes harboring likely causal common coding variants (reported from a recent study 17 with posterior probability of causal association >0.25 from genetics alone; Methods ), and 20 genes significant in a transcript association analysis with T2D 52 . Genes with likely causal coding variants demonstrated a significant set-level association relative to comparison gene sets ( p =8.8×10 −3 ) and to genes within the same loci ( p =0.028; Figure 2e ), even when we conditioned gene-level associations on all significant common variant signals.…”

A Powerful Framework for Integrating eQTL and GWAS Summary Data

“…A new gene-based association test called PrediXcan was recently proposed to integrate GWAS individual-level data with an eQTL dataset, alleviating the above two problems in boosting statistical power of GWAS and facilitating biological interpretation of GWAS discoveries (Gamazon et al 2015). It was extended to GWAS summary association data (Torres et al 2017). A similar approach, called transcriptome-wide association study (TWAS), was proposed by another group for GWAS individuallevel and summary data for one or more eQTL datasets (Gusev et al 2016).…”